And indeed everyone does seem to agree. Whether it's MG Siegler channelling Steve Ballmer, or John Gruber telling Guy English to let the Mac die quietly while iOS takes over, everybody agrees that the future of computing is mobile. Everybody except me.

The future of computing is mobile in the sense that more computing is being done on handheld devices. But that's because mobile computers have opened up more people to using computers than before, and opened up the set of circumstances in which it's possible to use computers. It's not so much that there has been a transfer of functions from desktop to mobile (although there has been some) but that there are things a mobile computer can be used for that a desktop or laptop can't. The reverse is equally true.

It's often said that only nerds need desktop computers. And that sounds plausible, but only if you define "nerd" loosely enough. Of course computer nerds need desktops. But then so do design nerds, if you take a look around. And so do audio nerds, and video nerds, and photography nerds, and typography nerds, and broadcasting nerds, and podcasting nerds, and data nerds, and science nerds, and engineering nerds, and finance nerds, and economy nerds, and policy nerds, and process management nerds, and project management nerds, and so on.

But at that point the word "nerd" doesn't mean what you thought it meant when you agreed that only nerds need desktop computers. It just means a specialist professional, and they're everywhere. Not only could I not do my job on an iPad, my mum couldn't do her job on an iPad either. And she knows almost nothing about computers. Her desktop is also just a platform for apps.

Now let me be clear about what I'm saying before you accuse me of technological conservatism. It's not that I can't see that a profound revolution has taken place in personal computing since the iPhone was released in 2007, or that I don't welcome it. It's that I think the revolution has been misrecognised. The revolution is touch, not mobility. It just looks like mobility because touch is what made mobile computing possible. With only indirect on-screen controls it's hard to use a computer held in two hands, let alone one. Touch is what makes iOS such a pleasure to use. (And lag is what makes Android feel inferior.)

Because you can't have the mobile computers we have today without responsive touchscreen technology, it's understandable that mobile computing took off so spectacularly when Apple led the way. Who doesn't want a computer this powerful and usable in their pocket?

But the rise of mobile computing today is a bit like the rise of home video in the 1980s. When home video players first became available, the breakthrough of being able to choose what you wanted to watch, when you wanted to watch it, felt so radical that people thought it would mean the end of cinema. How could cinema, with its programmed schedules, fixed screening times and distance from your living room possibly compete?

The answer, it turned out, is that size matters. The big screen experience offers something that simply can't be reproduced on a TV. Cinemas got bigger, the sound got better, the seats got more comfortable, the lines of sight were cleared. And today cinemas are still doing good business despite significant improvements in the size and quality of living-room TV sets and the availability of streaming movies on demand.

My hunch is that something similar is happening with mobile computing. Right now, while people are still discovering smartphones and tablets for the first time, the sheer convenience and usefulness of having a touchscreen computer available in a pocket or bag is so overwhelming that it seems as though the future could only be mobile. But the potential to revolutionise computing is much bigger than the mobile market.

As I walk around the different organisations that my work takes me into (many of which have enthusiastically embraced the iPad in various creative ways) it doesn't look like the desktop computer is going away. Just look at the on-screen real estate in the desktop set-ups of some of our favourite mobile computing evangelists. Do you think they're going to give up those generous workspaces anytime soon? I don't.

Size matters because artists like to work on large canvases. I can be remarkably productive on an iPhone when I'm out and about (I'm typing this sentence on one right now) but when I sit down in front of my desktop to undertake a significant piece of work I feel a sense of freedom and space that's just not there on even a ten inch screen.

There's room to breathe and stretch out, space to organise tools, freedom to arrange the workflow. Typing is faster. Data moves between applications with less friction. It takes a little longer to start up and set up (less now with an SSD) but once that's done I can achieve a level of productivity on a desktop that's simply not possible within the cramped confines of a small screen that invites interaction with just one app at a time.

Give it time, people say. But the year that mobile replaces the desktop is in danger of becoming like the year of desktop Linux: always just around the corner but never here today. And there are still significant obstacles to it.

For a start, the desktop is currently essential to building the mobile stack. In theory, we will one day be able to write software for mobile operating systems on mobile operating systems, but we can't do that right now. And we won't be able to until iOS gives developers the sort of access it currently denies its users with good reason. As Gruber has pointed out, the Mac is carrying iOS's baggage. But outsourcing complexity to more powerful tools doesn't make the need for more powerful tools go away.

So here's what I'm not saying. I'm not saying that desktops are better than mobile computers. I'm not saying that tomorrow's desktop computers should work the same way as today's. I'm not saying we need mouse pointers and user-facing filesystems forever.

What I am saying is that iOS won't obviate the need for a Mac until it is capable of doing the same substantial work as well a Mac. And an important part of what makes the Mac desktop so useful is its size and flexibility. Let's stop talking about desktop versus mobile computing. Let's talk instead about big-screen and small-screen computing, and recognise that we still want big-screen computing in the post-PC era.

The philosopher Daniel Dennett once said the biggest mistake in philosophy is to confuse a failure of the imagination for an insight into necessity. I think that looking at the current state of computing and concluding that the future is "mobile mobile mobile" represents a fantastic failure of the imagination.

Peggy Olson (the agency's bright young thing) eventually comes up with a campaign based around the idea that "you can't frame a phone call", the idea being that phone calls are intangible and disposable, while telegrams are treasurable objects that can mark a moment in time.

It's a great scene, because in the same moment that you recognise the cleverness of Peggy's idea, you know it's doomed to fail. Once everyone had telephones, the telegram was obsolete. There is nothing in the essential function the telegram performs that the telephone doesn't do better. The selling point that Peggy's campaign is based around is just a side effect of how the telegram works. It wasn't designed to deliver that feature.

Today, the printed newspaper is in the same position as the telegram was in the nineteen sixties. There is nothing in the essential function of printed news that the internet doesn't do better. You can get more news, to more people, more quickly, and at lower cost online. The only remaining advantages the newspaper has are things that are peripheral to its purpose. For instance, you can frame one.

In contrast to technologies like the telegram are technologies like the qwerty keyboard. While innovation has succeeded in making the telegram obsolete, there appears to be nothing it can do to improve upon the qwerty keyboard.

Famously, the qwerty keyboard was not invented to make typing easy. It was invented to reduce the frequency with which adjacent letters were typed, because the mechanisms of old-fashioned typewriters were prone to jam when that happened.

No modern keyboard uses a mechanism that suffers from this problem, but we persist in using a keyboard layout that makes typing more difficult because that's the layout our fingers already know how to use. So we expect our children to adopt a technology that's less than optimal because the cost of us learning a new layout is too great.

The defining feature of qwerty keyboard technologies is that their essential function is so indispensable that society standardises on the first successful implementation, but that implementation introduces features that prove to be undesirable in the long-term.

A good example of a more recently invented qwerty keyboard is Microsoft's Excel spreadsheet. We desperately need the thing that it does, but we are stuck with a terrible implementation. Because Microsoft snuffed out all competition at the height of its desktop monopoly, there is now little hope of a better alternative getting traction. Too much of the world's business logic and shared data is encoded and distributed in Excel for everyone to switch.

These two different kinds of technology — telegrams and qwerty keyboards — are useful categories for thinking about how post-PC devices have (and haven't) changed personal computing.

First, smartphones and tablets don't appear to have created many telegrams. The handheld gaming console is a possible candidate, as is perhaps, eventually, the mouse pointer. But the multitasking desktop doesn't appear to be going away, if only because, as Lukas Mathis points out, some tasks require the simultaneous use of several applications in combination to achieve a particular goal. Post-PC devices can be used at times and in places it simply wasn't possible to use computers before. But they haven't replaced PCs. They have displaced them in just those circumstances where smaller, more portable devices are better.

Second, some things that looked like telegrams turned out to be qwerty keyboards. For several years there has been something close to a consensus among the user experience community that the file system should disappear as a user application. And with good reason. The file system is arguably too big, too confusing, too labyrinthine, too difficult to navigate, too fragile, and too important to the computer's functional performance to expose to the user.

iOS took a bold step in not letting users see the file system and sandboxing applications so that they only had limited access to it. But at the same time, it didn't provide an alternative solution to the problems that the file system currently solves.

The mistake has been to assume that if users don't need to see the file system, then they don't need filing systems more generally. And yet the fact such systems existed for centuries before the invention of the computer suggests that people do need flexible, structured repositories for the information they use, produce and share. A computer's file system may not have ideal characteristics as a repository for digital information from a user's perspective, but it was rapidly adopted for that purpose precisely because something was needed to perform that role.

Where Apple has removed the file system on iOS, users have hastily bolted it back on in the form of Dropbox or other similar services. Dropbox improves on a basic file system by providing synching between devices; but it is interesting that the most popular file management service on iOS is the one that most closely resembles the file systems we already know how to use.

People need what the file system does, however badly the file system currently does it. And refusing to provide something that meets that need isn't going to make it go away.

The move feels desperate rather than planned; reactive rather than carefully executed. There has been a growing mood among App.net's subscribers that momentum has stalled, while its developers have been asking themselves whether writing software for the platform (it's designed to support a far broader range of services than Twitter) is economically viable.

The new free accounts are more limited than subscriber accounts — you can only follow a maximum of 40 people for a start. But App.net is clearly hoping that once people get on board with a free ticket they will eventually upgrade to first-class. I don't know whether that will work or not. At first glance it looks like a bait-and-switch. It has a different flavour to the straightforward offer that was presented to subscribers when the service first launched, which was membership of a premium ad-free social network in exchange for a uniform up-front fee.

Shortly before App.net announced its new membership policy, I had been discussing App.net pricing with Wessel van Rensburg. Wessel works for a social media agency in London and spends a lot of time thinking about these kinds of problems. He and his colleague Adriaan Pelzer have come up with an alternative pricing model for a subscriber-based social network that seems better designed to grow the number of paying members than App.net's approach.

Here's their idea. Start with a target number of subscribers and a target price, say a million users and £15 a year. The very first members are offered a price that is heavily discounted relative to the target price. You might offer the first one percent of subscribers annual membership at just one percent of the target price. The next one percent of subscribers get it at two percent, and so on. As the size of the membership grows the price increases towards the target price. Once you reach the target membership, which should also be the point at which the network achieves a critical mass, all new subscribers pay the target price.

There are various ways you could tweak this pricing structure, such as changing the rate at which the price increases, or starting with a thousand free accounts to seed the network, but the key elements are 1) that the price starts low and increases towards a ceiling relative to the number of sign-ups, and 2) that the pricing structure is established and clearly communicated from the start.

This idea has a number of things going for it. First, you are incentivizing people to subscribe right now rather than to wait and see how things develop. App.net's approach of starting with a high price and reducing it over time has a deflationary effect on the value of their subscriptions. If people think the price will fall in future, they will wait in order to pay less. But if the price being offered right now is the cheapest it's ever going to be, they are more likely to join at the first opportunity.

But more importantly, you are inviting subscribers to pay a price that better reflects the value of the social network at the point they become a member. While the network is small the price is low. As it grows the price increases.

This idea, that paywalls should charge users based on the value of the service to them at point they use it, has wider applications than just social networks. For some time I have been arguing with friends in the newspaper industry that news paywalls should take a similar approach to pricing as London's Oyster card.

For readers that don't know London, an Oyster card is an electronic ticket that can be used on pretty much all public transport in London. You add cash to the card at stations or online. But the killer feature is the way it responds to your spending. In London, before the Oyster card, you could buy one journey up-front or you could buy a Travelcard that let you take as many journeys as you wanted that day within a given area.

Most days it was obvious which option was better value, but often it wasn't. And if your plans changed you could end up spending more than necessary. The Oyster Card solves that problem. It starts by charging you for single journeys, up to the price of a Travelcard. But once you reach that price, you can travel at no extra cost for the rest of the day. So you always get the best value, without having to know what that is in advance.

What does this have to do with newspaper paywalls? Well take The Times paywall for example. (This is The Times of London not the NYT. Along with the Financial Times, it is one of the paywalls you encounter most frequently if you follow UK news.)

The cheapest route through the paywall is the basic digital subscription, which costs £3 per week and has a minimum subscription period of 52 weeks. So the cheapest way to read the article that led you to the website is to commit to spending £156. No-one is going to pay that in order to read just one article.

Perhaps The Times hopes you'll reach their website often enough to start thinking about subscribing. Maybe that happens to some people. But it misses the point, and it misses an opportunity. In the moment that you hit their paywall, the value of their website is just the value of the article you want to read.

The Times used to let you pay £1 for one day's online access. That's a bit better, but it's still too much. It may be a fair price for one day, but it's not a fair price for one article. Most people won't pay a pound for 800 words. I wouldn't. But I would pay 25p.

Here's how it could work. To read one article would cost 25p. Read four articles in one day and you'd get full access for the rest of the day. Pay for three days in one week and you'd get full access for the rest of the week.

Would that make me spend £3 a week on The Times? Not every week. But most weeks I'd spend something, which is a good deal more than I spend now. And some days I'd earn the freedom to find more articles and authors that I valued reading. Pricing access in this way would open up a stream of revenue from the long tail of people who are happy to pay something, but don't want to pay for everything.

Online services should offer a price that reflects the value of their product at the point their customers pay for it. Let's try that before we decide that paywalls are destined to fail.

It seems insane that investors would prefer a company that earns $0.21 per share to one that earns $13.81, when AAPL is less than twice the price of AMZN. And especially when Apple has around $137 billion of cash in the bank, which is more than the total value of Amazon's shares in circulation.

Of course not everyone is impressed with Amazon's performance, and some experienced venture capitalists are still willing to make the case for Apple. But the share price of both companies today remains roughly where it was in the aftermath of January's earnings calls.

Some people have pointed out that although Apple's revenue and profit have both increased, revenue has increased more than profit, so the proporiton of Apple's revenue that is profit has gone down. But if tighter margins were the only reason for the market to react in the way that it did, then Amazon's shares would have been falling over the last three years.

So is the market wrong? The market can be wrong. As Keynes famously observed, the market can be wrong longer than you can stay solvent. And it might be wrong about Apple and Amazon right now. But there is method in its madness. And it's to do with the way the market thinks about risks and returns.

The prism through which the market looks at financial assets is Portfolio Theory, which attempts to reduce the risk associated with a given return (or increase the return associated with a given risk) by ensuring that the risks within a portfolio of assets tend to cancel out.

According to the theory, you should invest in a set of assests whose individual returns are broadly uncorrelated, so if one asset performs worse than expected, that will be offset by another asset performing better than expected. In plain English, it's the simple idea that you shouldn't put all your eggs in one basket.

Portfolio Theory is a way of thinking about the risks and returns of a whole investment portfolio rather than the quality of an individual asset. But the principles at work in Portfolio Theory can apply to the risks and returns associated with a particular business.

Tim Cook once famously told an audience at Goldman Sachs that Apple could fit its entire family of products on the table they were sitting around. But where Apple's admirers see a sharp focus, and its customers see an elegant simplicity, the market sees a troubling lack of diversity.

In the last quarter over half (56%) of Apple's revenue came from the iPhone, and three quarters (76%) came from the iPhone and the iPad combined. These are products that most customers will buy once every 18 to 24 months, which means Apple is only one or two purchasing decisions away from losing each customer from the group that contributes most of its revenue. Apple's lack of diversity may be the reason the few things it produces are so good, but it's a potential Achilles' heel.

Contrast this with Amazon, from whom people buy so many different things, so frequently, the company can sell “all you can eat” shipping as an annual subscription. Amazon doesn't suffer much if its customers go elsewhere for occasional boutique purchases, as long as Amazon is their first-choice online retailer for most things.

None of this matters in the short term, when Apple is making $13.1 billion a quarter in profits, compared with Amazon's $97 million. But investors calculate the value of a financial asset as the present value of its future returns, and that depends on what sort of a future you think a company has.

The breadth of Amazon's inventory, the scale and efficiency of its distribution network, and the thinness of it's margins makes it very difficult to unseat as the incumbent online retailer. But if the iPhone falls, then Apple falls. (Or at least, it falls as a company that can compete on market cap with Exxon Mobil.)

That's why the market panics at any hint demand for the iPhone might be weakening. It's also why iPhone releases are becoming an increasingly high stakes tightrope walk for Apple. Do something too radical and risk alienating customers, do something too incremental and risk them looking for innovation elsewhere.

So back to where we started: is the market right to shrug off Amazon's tiny profits and worry how far Apple's earnings have to fall?

Not necessarily. Apple could dominate personal computing for the next two decades just as Microsoft did for the last two. And with the increasing digitisation of content, someone may find a way to unseat Amazon more quickly than anyone currently imagines is possible. But the market just lengthened the odds on both of those outcomes. And if you think the market is wrong, you can always place your bets.

Disclosure: I own no shares in either Apple or Amazon. I only invest in funds that track broad indices. (Because buying shares is not like buying a computer.) None of the above should be taken as investment advice.

That another record quarter of revenue and profit for Apple could lead to a 10% fall in its share price does indeed raise questions about what the market expects from the company and whether it's being rational. But the market reaction is interesting in another way, because it highlights irrationality on the part of Apple's supporters too.

The previous week both Gruber and Marco had linked to a story from last year that suggested Apple's low share price just before the earnings call was nothing to do with market irrationality but was instead due to stock manipulation. The financial details of the story were complicated, but it implied that analysts were using misinformation to suppress the price of Apple's shares ahead of an earnings call that was likely to trigger a bull run.

In the event, the true information revealed in that earnings call did more damage to Apple's share price than any rumours beforehand. With hindsight, it now looks less like the stock was being manipulated and more like the market was correctly anticipating what it would regard as bad news. Whether it was bad news or not is an interesting question, but either way the stock manipulation theory seems much less plausible now, because there's no reason to assume Apple's pre-earnings share price was artificially low.

“Well so what?” you might ask. I think it matters, because those of us who like and use Apple products are quick to spot prejudice when it appears in criticism of Apple but are perhaps less good at recognising our own cognitive biases when they appear in defence and praise of the company. Marco himself recently wrote a piece about anti-Apple anger, in which he pointed out that:

Android fans are unusually quick to fill the comment box with rage on articles that mention anything positive about Apple or its products. The reverse – Apple fans leaving angry comments on pro-Android articles – is almost completely absent from the sites I’ve seen, including sites like The Verge that have many readers in both camps.

That's true enough. But that doesn't mean Apple fans don't have their own prejudices. They're just manifested in a different way.

In as much as there really is something you could call anti-Apple anger there is also something you could call pro-Apple paranoia. Apple has always tended to think of itself as an underdog and an outsider, and its oldest and most loyal customers have grown used to seeing it as a company under siege – from ruthless competitors, greedy markets, and a hostile press that seems determined to misunderstand it.

Those things may be real enough, but that doesn't mean we have to believe in conspiracy theories.

Some people have criticised this visual design on the grounds that it makes the app harder to use. And if you are not familiar with how a tape-deck works, that's almost certainly the case.

If you have used an old-school tape-deck, some aspects of the design do provide useful visual feedback. The amount of tape on each reel accurately represents the current position within the recording, while the buttons for skipping backwards and forwards cause the reels to spin the tape the appropriate distance in each direction, which neatly illustrates their function.

All of this may be lost on someone for whom a tape-deck is as unfamiliar as a gramophone. But the problem with this interface isn't just that it makes the app harder for some people to use. It's that it's inappropriate and unnecessary.

Podcasts have never been consumed on tape. Podcasting only became possible with the advent of digital distribution online. So the very existence of podcasting as a medium demonstrates that the technology represented in Podcasts' interface is obsolete. In that sense the app's design is not just skeuomorphic. It's anachronistic.

Am I being too literal? Podcasts are just audio, and tape is audio media, so the visual metaphor of a tape-deck works fine in that sense, right? But if that's the justification, why is the tape-deck interface only used in the Podcasts app and not in the main music player? Podcasts' interface was designed to be distinctive, but it's distinctive in a way that doesn't reflect what's distinct about podcasting.

This kind of conceptual incoherence matters. It's why using a floppy disk as a ‘Save’ icon is so jarring, while few people notice the symbolism when an envelope is used to represent an email. No-one saves data to floppy disks anymore, but a letter is still a pretty good physical analogue to an email.

My favourite example of conceptual incoherence in interface design is the Vatican website, which uses the most advanced publishing technology in the world to produce pages that look like medieval scrolls. Nothing says you're uneasy with the modern world like rendering hypertext as parchment.

Does all skeuomorphic design suffer from this sort of incoherence? No. But I think it often does in the field of software interfaces. At some point the similarity between the skeuomorphic interface and the technology it represents breaks down. Even the apparently straightforward metaphor of a letter for an email breaks down in some important ways – most notably with regard to privacy. And even if a skeuomorphic interface remains perfectly compatible with what it controls, strict adherence to modelling the properties of the original technology means not doing new things that simply weren't possible in the past.

Skeuomorphic interface design leverages familiarity with an old technology to sidestep the problem of deciding what the best interface for a new technology should be. It draws on the wealth of the past, but only by short-changing the future.

At one point I thought she'd finished experimenting with it and I started to say something just as she pushed Siri's microphone button, “beep-beep”. As soon as she realised I was talking, she tapped it again to stop Siri listening. But Siri must have heard something, because after a short pause it cheerfully announced that it was calling the last person in my address book that I would want to disturb with an unexpected late night phone call.

The person Siri had decided to call isn't someone I know very well, but he's pretty important in my professional universe. I had his number for obscure work-related reasons and I had been given it on the understanding that it should only be used in certain circumstances, which didn't include experimenting with gadgets.

I quickly grabbed the phone, hung up the call, and immediately disabled Siri. In that moment I decided that Siri wasn't just unreliable, it was potentially dangerous. I didn't want to give the power to make calls to a cloth-eared daemon that could interpret any garbled statement as an instruction to call a random contact. From that point on Siri was retired. And when Apple replaced my phone because of a failing battery, I declined the invitation to turn Siri back on.

Until last weekend.

I don't know why I decided to give Siri a second chance. I guess I started to wonder if I had been too hasty. Siri was always expected to improve as millions of users feed it the training data it needs to learn. Perhaps Siri had become smarter and less error-prone. And with iOS 6 coming out soon, I was in the mood for trying things out.

So I enabled Siri and opened it up. The software immediately offered a list of example questions that Siri would be able to answer. Apple doesn't yet have a partner in the UK for Siri's business search, so I looked for a question Siri should be able to answer in the UK. The first one I noticed: “What's the weather for today?”

Beep-beep. “What's the weather for today?”, I asked. And this is what came back.

Now I react to a response like this on two levels. The part of me that knows something about computer science understands that speech recognition and natural language processing are two of the hard problems of artificial intelligence, and Siri is trying to do both of these things in performing one task.

It's hard to identify individual words in a waveform of speech. And real-world utterances are riddled with ambiguities that humans resolve through an understanding of context and shared knowledge, both of which computers lack. So I know it's unfair to slam Apple for the fact that Siri makes these kinds of mistakes, because it's not an implementation failure. It's just the current state of the art.

On the other hand, you can see how a technologically naive user would experience an error like this. Siri invites you to ask it a particular question in order to show you how it works. And then it doesn't work. It's like having a magician proudly declare "This is your card!", and it's not your card. An experience like this can really undermine the perceived value of the software.

The difficulty for Apple is that Siri is a fundamentally different kind of software to the kind they excel at producing. It's not something you can release into the world near-perfect and fully-formed. To reach its full potential Siri needs to make mistakes. But every mistake it makes comes at a cost to someone somewhere using the software. And that's not something Apple's customers are used to. The challenge for Apple is to make Siri useful enough for people to keep using it, without setting expectations too high. And I'm not sure they've got the balance right.

I decided to give Siri a second chance. But with first impressions like these, I wonder how many people do the same.

It doesn't show the absolute number of searches, but an index that represents the number of searches at each point in time relative to the average number over the period. But given that this chart covers the entire period during which Facebook has existed, it's a fair way to compare the frequency of searches for “Facebook” over time.

Now these numbers aren't registrations, or page views, or time spent on Facebook, so they're not a direct measure of how much or how many people are using the service. But I think they may be a good proxy for the number of visits to Facebook's website. Why? Because, as the Facebook login page-rank incident demonstrated two years ago, many Facebook users navigate to the website by typing the word “Facebook” into Google.

If that's right (and assuming that the proportion of people that get to Facebook via Google remains the same), then visits to Facebook's website have been flat for over a year. That doesn't necessarily mean Facebook has stopped growing. At least some of the levelling off on the web could be due to people increasingly interacting with Facebook through mobile apps, which bypass the need for a Google search.

But either way, if visits to Facebook's website have flattened out, that has significant implications for Facebook as a business, because the website has so far been the main focus of the company's efforts to raise revenue through ads and apps. And even if Facebook is still growing on mobile devices, the company has only recently started using its mobile interface as an advertising platform, and is limited in the ways it can use it as an apps platform, given the constraints that are imposed on mobile applications. (Consider that when a third-party developer writes a native iOS version of their Facebook app, they hand some control over that part of Facebook's ecosystem to Apple.)

Still, it's not as though searches for “Facebook” are actually declining anywhere, right?

But when it comes to the content industries, hosts John Siracusa, Marco Arment and Dan Benjamin often regret the fact that there are no TV companies doing the Apple thing – allowing customers to buy just the TV programmes they want to watch, on demand and without advertising – because advertisers are willing to pay more than audiences for content.

This made me wonder if it's possible to work out just how much advertisers pay for our time in front of the TV, to see how it compares with what we might be willing to pay to watch the same shows ad-free. Looking at the available data, I think it's possible to come up with a ballpark estimate. And the amount that advertisers pay on average seems surprisingly low.

Working out the average TV advertising revenue per viewer is not straightforward. Different TV channels have different sources of revenue and advertising plays a different role for each one, depending on whether they have paying subscribers or are receiving some form of public subsidy. But there is one channel in the UK that is both representative of the national TV advertising market and that allows for a fairly simple estimate of its advertising revenue in terms of the amount it generates per viewer hour.

ITV1 is the UK's largest national free-to-air commercial TV channel. It has the second largest audience of any TV channel after BBC1 and doesn't receive much in the way of public subsidy. Data on ITV1's national audience is published weekly by the Broadcasters' Audience Research Board. This includes both the total number of viewers each week and the average weekly viewing time per viewer. Together they can be used to calculate the total number of viewer hours each week. Meanwhile, ITV1's total advertising revenue each year is published by the media regulator Ofcom in its annual Communications Market Report.

Based on the most recent data, ITV1's total advertising revenue in 2011 was around £1.3 billion, while total viewer hours were around 9.5 billion. Dividing the revenue by the viewer hours leads to average revenue per viewer hour of just under 14 pence (22 cents).

Now there are a couple of things to note about this estimate. First, it's an estimate for just one TV channel in one country. Revenue per viewer hour could be larger or smaller in other countries depending on the size of the national TV audience, the size of the advertising market, and the nature of competition in the TV industry. Second, it's an average figure across all time spent in front of the TV. Revenue per viewer hour will vary considerably depending on the programmes those hours are spent viewing. An hour of prime time drama will generate more revenue per viewer than an hour of people talking on a sofa in the middle of the day.

But even if you assume that time spent watching the highest quality programmes is worth many times the average, while time spent on low-quality shows is worth just a fraction of it, the amount of advertising revenue generated per viewer hour doesn't seem unaffordable to an individual viewer who is motivated to get rid of the ads. So how is it that we, the audience, are being outbid at prices many of us would be wiling to pay?

I'm not an economist so I lack the vocabulary to properly describe the problem in those terms, but I think I can see the shape of it. The answer, I think, is to do with the difference between paying nothing and paying anything at all.

Start by imagining what would happen if a TV channel tried to switch its entire revenue stream from its advertisers to its audience. There would be no more ads, instead viewers would pay for TV at the same rates per viewer hour that advertisers are effectively paying now.

The law of demand says that the higher the price for a particular good, the smaller the number of people willing to pay for it. So as viewers we can't in practice pay the same as advertisers do for the time we spend watching TV, because paying directly for TV would change our viewing behaviour. Knowing that the meter was running when the TV was on would force us to ask whether the programmes we were watching were really worth the price. We might enjoy the TV we watched more, but on average we would watch less of it. And that would mean the TV channel would need to charge higher prices per viewer hour to make the same revenue. And higher prices would further reduce viewing, and so on.

Perhaps somewhere down that road there is a point where the numbers balance out, and the TV company makes a decent profit producing less TV for fewer people. But the average price per viewer would be much higher than 14 pence an hour; and the total profit for the TV company would almost certainly be smaller.

So this is how advertisers outbid audiences: they outbid them in total. As long as the total amount of money that advertisers are willing to pay is more than the total amount the audience would pay for the same viewing behaviour, advertisers will outbid the audience. Many viewers may be willing to pay a higher price than advertisers currently do for the TV they watch, but the audience as a whole must offer more for the same content if they don't want to be outbid. So viewers who would rather pay can't just pay for themselves at the advertisers' rates, they also have to cover the cost of reducing the audience to just the paying customers.

TV companies find it hard to let viewers opt out of advertising individually for a similar reason – it reduces the value of what the advertisers are paying for. Viewers would need to pay a premium to opt out of advertising. And the more attractive the viewer is to advertisers, the higher that premium would be.

The obstacle to ad-free television is the assymetry in the objectives of advertisers and audiences. Audiences want the TV they watch to be the best possible experience, while advertisers want you to watch as much TV as possible. You may be willing to pay more to watch fewer hours of better TV. But unless everyone is willing to make the same trade-off at the same prices, it's unlikely to happen soon.

Note on sources: Total advertising revenue for ITV1 in 2011 is taken from Figure 2.26 on page 140 of Ofcom's Communications Market Report 2012 (ITV1 plus ITV Breakfast). Total viewer hours have been calculated using BARB's Weekly Total Viewing Summary. The spreadsheet is here.

A fireball is a pretty good metaphor for this – a sudden explosion of traffic that gives way to a sustained blaze that slowly burns through a good chunk of your bandwidth over the next 16 hours.

The site stayed up. And that was a relief. It uses a custom blogging engine that could generously be described as experimental. To call it a CMS would be an overstatement. The codebase is still small and it lacks features on the back-end. But I like it and feel like I've got the fundamentals right. It will grow with the site.

Of course I would like to believe that surviving the fireball was entirely due to my superlative web development skills. But to be honest I think the timing of Gruber's post helped a lot. As best I can tell it went up at around 2:30 GMT, when most of Europe was fast asleep and the East Coast was starting to go to bed. That must have spread the traffic out. And looking at that chart, it's really all about the first five minutes.